REVIEW G-Protein-Coupled Receptors: From Classical Modes of Modulation to Allosteric Mechanisms Thomas M. Bridges†,* and Craig W. Lindsley†,‡,* †
Vanderbilt Program in Drug Discovery, Department of Pharmacology, Vanderbilt Medical Center, Nashville, Tennessee 37232 and Department of Chemistry, Vanderbilt University, Nashville, Tennessee 37235
‡
A B S T R A C T Heterotrimeric G-protein-coupled receptors (GPCRs) represent a large protein family responsible for mediating extracellular to intracellular signaling within a broad range of physiological contexts. Various conventional models have been used to describe their interactions with ligands and G-proteins. In recent years, however, numerous novel ligand⫺receptor interactions not adequately addressed by classical receptor theory have been recognized. In addition to traditional orthosteric ligands, many GPCRs can bind allosteric ligands that modulate receptor activity by interacting with distinct or overlapping receptor sites. Such ligands include positive allosteric modulators, which have become the focus of pharmaceutical drug discovery programs and have gained the attention of a growing body of basic and translational researchers within the academic community. Here, we review the fundamental aspects of allosteric GPCR modulation by smallmolecule ligands, with particular focus on the emerging position of positive allosteric modulators in modern drug discovery.
*Corresponding authors,
[email protected],
[email protected].
Received for review May 19, 2008 and accepted June 27, 2008. Published online July 25, 2008 10.1021/cb800116f CCC: $40.75 © 2008 American Chemical Society
530
ACS CHEMICAL BIOLOGY • VOL.3 NO.9
G
-protein-coupled receptors (GPCRs) represent nearly half of the current market for therapeutic agents, constitute annual revenues in excess of $40 billion, and remain a primary focus of many biomedical research and pharmaceutical drug discovery programs. Despite efforts by multiple laboratories over several decades, the crystal structures of only two GPCRs (the bovine rhodopsin and human 2-adrenergic receptors) have been solved definitively (1–4). Since the initial report on rhodopsin, many homology models for diverse GPCRs were constructed based on this crystal structure. The first report from Koblika and coworkers (2) described crystallization of the 2adrenergic receptor in a lipid environment bound to an inverse agonist affording low resolution (3.4−3.7 Å). Subsequent reports with engineered human 2adrenergic receptor refined resolution down to 2.4 Å, which demonstrated significant differences between rhodopsin and the 2-adrenergic receptor, highlighting the challenges and cautions of employing rhodopsin alone as a template model for such a large receptor family (3, 4). These new crystal structures provide exciting opportunities for analyzing orthosteric and allosteric binding sites as well as redefining GPCR structure. However, much is known about the basic structure and function of many GPCRs because of decades of biochemical, genetic, imaging-based, and molecular pharmacological research. Here, we first review the fundamental aspects of GPCRs in terms of their general structure and functional characteristics. GPCRs comprise a diverse family of integral membrane proteins that are responsible for conveying extracellular signals to the inside of the cell via interactions with intracellular heterotrimeric G proteins, which in turn affect enzymes, ion channels, and other intracellular messengers. Nearly a thousand GPCRs exist, mediating www.acschemicalbiology.org
REVIEW a host of molecular physiological functions by serv• Light ing as receptors for hormones, neurotransmitters, • Ca2+ cytokines, lipids, small molecules, and various • Odorants/pheromones • Small molecules sensory signals (such as light and odors), to name • Proteins a few. All GPCRs possess seven transmembrane helices, three extracellular loops, and three intraNH2 e1 e3 cellular loops, with an extracellular N-terminal tail e2 and an intracellular C-terminal tail (Figure 1). The heptahelical transmembrane domain is largely hyTransmembrane heptahelical domain Effectors drophobic, whereas the extracellular (e1⫺e3) and (7TM domain) intracellular (i1⫺i3) segments, or loops, are generi2 COOH α γ i1 ally hydrophilic, as would be anticipated for amino i3 β acids exposed to the phospholipid-rich memG protein Intracellular brane and the water-rich environments, respecmessengers tively. The seven transmembrane helices are each ⬃24 amino acids long, while the C- and N-terminal Figure 1. Representative structure of a generic GPCR. GPCRs all have a common core comtails as well as the loops can vary widely in length posed of seven transmembrane helices (the 7TM domain composed of TM-IⴚTM-VII) with with up to hundreds of amino acids. For example, an extracellular N-terminal domain and an intracellular C-terminal domain. The TMs are connected by three extracellular loops (e1ⴚe3) and three intracellular loops (i1ⴚi3). The the metabotropic glutamate receptors possess GPCR receives an extracellular stimulus (light, calcium, odorants, pheromones, small molN-terminal segments in the 600 amino acid range ecules, proteins) that induces a conformational change in the receptor that either facili(5). On the basis of sequence homology and func- tates or inhibits the coupling of the receptor to a G-protein, composed of ␣-, -, and tional roles, GPCRs can be divided into five or ␥-subunits. The G-protein, in turn, interacts with a diverse group of effectors that control intracellular messengers. sometimes six families, although the most common division is into three main families (or classes): A, B, and C (Figure 2). The families are binding to the receptor alters the equilibrium, with agoreadily distinguished by comparing their amino acid se- nists shifting it toward the R* state, inverse agonists quences, wherein Family B is characterized by a large shifting it toward the R state, and antagonists preventextracellular loop and Family C has a large, bilobed extracellular Venus-flytrap-like doH 2N NH2 main. A second major difference between the families concerns the location of the orOBD OBD NH2 thosteric binding site and the nature of the orthosteric ligand. As shown in Figure 2, the orthosteric binding domain (OBD) of Family A GPCRs is located with the OBD 7TM domain, whereas the OBD is located in the large extracellular loop within FamCOOH COOH COOH ily B and within the extracellular VenusFamily A Family B Family C flytrap-like domain in Family C. According to traditional two-state modH O HOOC COOH (Pyr)-PLPDCCRQKTCSCRYLE els of receptor theory, GPCRs can be conN+ H2N-LTLIGAAHNGAGHLL O NH2 ceptualized as operating in equilibrium be1, acetylcholine 2, orexin A 3, glutamate tween two functional conformations, an active (R*) and inactive (R) state (6). In the R* state, the receptor has higher affinity for Figure 2. Representative structures of the three families of GPCRs, Family A, Family B, and Family C. Note the location of the OBD varies for the families, as does the structure of the extracellular doG-proteins, which normally exist apart main. The nature of the orthosteric ligand also varies across GPCR families. For Family A, a protofrom the receptor as a GDP-bound G␣␥ typical native agonist is acetylcholine (1), for Family B, a large 33-amino acid peptide such as orexin A (2), and for Family C, glutamate (3) is a representative native ligand. heterotrimer in their inactive form. Ligand www.acschemicalbiology.org
VOL.3 NO.9 • 530–541 • 2008
531
TABLE 1. Abbreviated list of the primary
downstream effectors of each of the main G-protein subunits G-protein subunit
Effector target(s)
G␥
Ion channels, GIRK, PI3K, phospholipases, adenylyl cyclase Adenylyl cyclase (⫹cAMP), Na⫹ and Cl⫺ channels Adenylyl cyclase (⫺cAMP), K⫹ and Cl⫺ channels, phospholipases Phospholipase C
G␣s G␣i/o G␣q/11
ing other ligands (such as endogenous agonists) from binding without altering the basal R*:R equilibrium. Upon receptor activation, the GDP-bound G-protein interacts with the intracellular face and C-terminus of the receptor, inducing GDP to GTP exchange on the G␣ subunit and concurrent dissociation of the G␣ and G␥ subunits. The now active GTP-G␣ and G␥ subunits then bind to their respective downstream effectors, which include kinases, phosphatases, small GTPases, integral
GPCR Gβγ βγ Effector
Agonist
GαGDP
Gβγ βγ
GPCR GαGTP
GαGDP
Gβγ βγ
Effector GαGTP
Gα αGTP
GAP Effector
Effector Gβγ βγ
Effector
Figure 3. Generalized diagram of the G-protein cycle. Upon agonist activation of the receptor, GTP binds to the G␣ subunit, displacing GDP, which causes dissociation of the protein complex from the receptor, allowing respective effector activation by G␣-GTP and G␥. GAPs then bind G␣ and accelerate hydrolysis of GTP to GDP, which deactivates G␣ and causes disengagement of the effector. Finally, G␣ reassociates with G␥, marking cycle completion.
532
VOL.3 NO.9 • 530–541 • 2008
BRIDGES AND LINDSLEY
membrane proteins, and a multitude of additional targets and signaling cascades. These downstream effectors exist in complex regulatory networks that control cellular functions such as movement, metabolism, membrane potential, neurotransmitter release, and gene expression. Although the simple two-state model has historically provided a highly useful framework for describing and conceptualizing receptor activity, it has been greatly expanded and modified over time as more detailed and complex ligand⫺receptor phenomena have emerged. As discussed in more depth below, current receptor theory models have grown to encompass a vast number of pharmacological scenarios, and these models currently play instrumental roles in the modern understanding and quantification of ligand⫺receptor interactions. The specific effectors influenced by a given GPCR depend on the type of G-protein that the receptor activates. There are many types of G␣, G, and G␥ subunits, allowing for diverse combinations, although the most commonly used simple categorization of GPCRs is by designation of coupling to either G␣q, G␣i, or G␣s (7). The mutual effector for both G␣i and G␣s is adenylyl cyclase (AC), which resides on the inner leaflet of the plasma membrane and generates cyclic-AMP in response to stimulation or inhibition by G␣s and G␣i, respectively. The primary effector for G␣q by contrast is phospholipase C, a membrane-bound enzyme that converts phosphatidylinositol-4,5-bisphosphate into diacylglycerol and inositol-1,4,5-trisphosphate. Table 1 provides a limited list of the effectors for each G␣ subunit. Following effector binding, the GTP-G␣ subunit hydrolyzes its ␥-phosphate by augmentation of its intrinsic GTPase activity via binding of GTPase activating proteins (GAPs), resulting in conversion to GDP-G␣. This GDP-bound form possesses higher affinity for its G␥ subunit partner, which causes reformation of the inactive heterotrimer, marking completion of the G-protein activation cycle. Figure 3 depicts the cycle in this simplified form. Many accessory proteins and lipids are also involved in regulating G-proteins, which play important roles in controlling the G-protein cycle (7). Classical GPCR ligands modulate receptor signaling by directly stimulating a receptor response (agonism), blocking the binding of the native agonist (competitive antagonism), or blocking constitutive activity (inverse agonism) of the GPCR. Classical ligands exert their modulatory effects by interacting with the orthosteric www.acschemicalbiology.org
REVIEW binding site and have typically been characterized a using orthosteric radioligand binding methods. γ-Aminobutyric acid type B receptors (GABABRs) GPCR signaling is also substantially influenced by receptor expression, desensitization, and internalOH N N N N ization in response to binding by different ligands Cl N Cl N H H and within various cellular contexts (8). A major isHO sue for GPCR agonist drugs that require chronic administration is receptor desensitization, downCF3 regulation, and/or internalization over time. By 1, GS39783 3, GS39783-12a 2, CGP7930 effectively turning the receptor “on” in an unnatuEC50 = 3.1 µM EC50 = 5 µM EC50 = 5.3 µM 8-fold 127% 6-fold ral context, the receptor may lose sensitivity to the agonist and/or be internalized into the cell, where it b Adenosine receptors (ARs - type A1A and A3A) is no longer available to receive extracellular stimuli. Perhaps the most widely used example of NH3+I− N S N such regulation is the desensitization and internalS NH2 N ization of the 2-adrenergic receptor (2-AR) in reH3 C N HN O sponse to chronic activation by the agonist isoproO NH terenol (9). This involves binding of the 2-AR by N N -arrestin, a widely expressed cytoplasmic regulaH CF3 tory protein that binds GPCRs following receptor A3A A1A OCH OMe MeO 3 phosphorylation by GPCR kinases. Once bound, 7, DU124183 5, IDTA-3x 6, VUF5455 4, PD81,723 -arrestin blocks interaction between the receptor EC50 = 210 nM EC50 = 1.0 µM EC50 = 72.5 nM EC50 = 10 µM and G-proteins, uncoupling the signaling mecha130% 183% 3-fold 492% nism. -arrestin can also induce receptor internalc ization, reducing the pool of receptors available to Muscarinic acetylcholine receptors (M4) ligand binding and thus diminishing signaling to Cl downstream effector pathways (10, 11). Although NH2 Cl HN the role of arrestins in receptor internalization and NH2 HN trafficking appears somewhat ubiquitous, there are S O H3CO N arrestin-independent mechanisms as well. FurtherS O N more, tight coupling to G-proteins is in some cases 8b, VU100010 8a, LY2033928 absent or secondary to signaling via alternative EC50 = 400 nM 47-fold pathways (12). An example of such non-G-proteinmediated signaling is the direct activation of ERK2 by the 2-AR following arrestin binding (13, 14). The Figure 4. Representative PAMs of Family A GPCRs. emerging details and importance of such noncacally common and relevant of these ligands are the ennonical signaling by GPCRs are areas of growing interdogenous agonists. est and investigational focus because such signaling is The location and mechanism of agonist-induced actilikely to underlie many interesting and underapprecivation differ considerably from one class of receptors ated context-specific physiological mechanisms and to another; however, common themes have emerged. functions. Orthosteric versus Allosteric Ligands. All GPCRs pos- For all GPCRs, binding and activation by an orthosteric sess a distinctive binding site for their respective endo- agonist ultimately results in a structural rearrangement genous ligand(s) that is known as the orthosteric site. Li- of the receptor that results in increased affinity for G-proteins. In the case of most biogenic amines, nuclegands that bind to this site are considered classical or osides, and lipid moieties, ligand binding occurs within traditional orthosteric ligands. This group includes the hydrophobic core of the receptor. By contrast, smallsmall-molecule agonists, partial agonists, antagonists, peptide hormones bind to the core, extracellular loops, and inverse agonists; in general, the most physiologiwww.acschemicalbiology.org
VOL.3 NO.9 • 530–541 • 2008
533
Metabotropic glutamate receptor (mGluRs 1, 2, 4, 5) mGluR1 NAMs
mGluR1 PAM O
O
O N H
mGluR2 PAMs OH O
O O MeO
N
9, Ro-674853 EC50 = 10 nM
N
O
S
CN
O
12 EC50 = 890 nM 68%
11, CBiPES EC50 = 94 nM 3-fold
10, JNJ 16259685 IC50 = 20 nM
mGluR5 PAMs
mGluR4 PAM HO
N
N O S O
Cl
N O O
N
F
O
N
F
O N
HN
HN
N
OH
N
H N
CN O
O
14, DFB EC50 = 3.1 µM 3-fold
13, (-)-PHCCC EC50 = 4.1 µM 4-fold
15, CPPHA EC50 = 280 nM 9-fold
16, CDPPB EC50 = 26 nM 5-fold
mGluR5 NAMs S N
O NH
N N
N
O N H
Cl
N
17, MPEP
18, MTEP
19, fenobam
Figure 5. Representative PAMs and NAMs of Family C GPCRs.
and the N-terminal segment of their respective GPCR. Larger proteins, and glycoproteins in particular, generally bind to the N-terminal tail, which then moves down to establish ligand-loop interactions to activate the receptor. Similarly, many neurotransmitters including GABA and glutamate bind to the large N-terminal tails found on metabotropic class C neurotransmitter receptors, causing a conformational change that brings the N-terminus down to the transmembrane domain to induce activation. The precise structural changes involved in transducing ligand-binding to G-protein-binding are diverse and complex, but they can be generalized to involve the disturbance of ionic interactions between transmembrane helices three and six (15, 16). This facilitates binding of the G-protein predominately by helices two and three as well as the C-terminal tail (7). In the case of drug discovery, much of the field’s history has revolved around the identification and study of small molecules that act as orthosteric ligands at a given target receptor to elicit a pharmacological effect. These compounds compete with the endogenous ligand(s) and thus preclude simultaneous occupation of the receptor by both molecules. Figure 2 depicts the orthosteric ligand binding site for each of the three main GPCR classes along with representative endogenous (orthosteric) agonists for each class. Classical drug discov534
VOL.3 NO.9 • 530–541 • 2008
BRIDGES AND LINDSLEY
ery focused on radioligand binding assays to identify “hits”, which were, by default, orthosteric ligands. In addition to orthosteric sites, many GPCRs have been found to possess allosteric (Greek, “other site”) binding sites that are spatially and often functionally distinct (17–19). The presence of allosteric sites allows for numerous additional ligand⫺receptor interactions beyond those associated with the orthosteric site. Allosteric agonists, antagonists, and inverse agonists for a given GPCR will bind to the allosteric site and induce a similar effect as their orthosteric relatives. Beyond such types of ligands, allosteric modulators bind to an allosteric site where they stabilize a receptor conformation and equilibrium shift that increases or decreases the affinity and/or efficacy of an orthosteric agonist at the receptor, without activating the receptor on its own (17– 19). The modulator lacks intrinsic agonist or inverse agonist activity, and thus increased or decreased signaling via the receptor occurs only in the presence of an orthosteric agonist. Such ligands are often respectively termed positive allosteric modulators (PAMs) and negative allosteric modulators (NAMs), and examples of each are highlighted in Figure 4 and Figure 5 for Families A and C, respectively (17–19). With the evolution of high-throughput screening (HTS) and functional assays, scientists were able to identify molecules that afwww.acschemicalbiology.org
REVIEW
www.acschemicalbiology.org
EC20 agonist plus 10 µM positive allosteric modulator
11000 Fluorescence change (counts)
fect the function of GPCRs irrespective of binding mode. Figure 6 highlights how a functional HTS screen, typically Ca2⫹-mobilization measured by fluorescence with a calcium-sensitive dye, is designed to identify positive allosteric modulators (20). The sometimes confusing concept of allosteric modulation by a potentiator is often best visualized by looking at the effect of a PAM in two simple cell-based pharmacological assays: a full concentration response curve (CRC) of an orthosteric agonist in the presence of increasing concentrations of potentiator (Figure 7, panel a) and the full CRC of a potentiator in the presence of a fixed low concentration of orthosteric agonist (Figure 7, panel b). Together, these assays demonstrate a shifting of the orthosteric agonist potency (and/or efficacy, depending on the potentiator), which can be translated into a quantitative “fold over baseline” or “fold-shift” as a measure of potentiation. Although it is not yet well understood what degree of fold-shift is required for significant effects in vivo, some published examples demonstrate in vivo significance with as little as 3⫺6-fold leftward shift of the orthosteric agonist CRC (21). As one would expect, the specific physiological or pathological context is likely to determine the degree of potentiation required to observe in vivo effects, which cannot be easily generalized. Furthermore, the exact mechanism(s) involved in mediating the effects of a potentiator are likewise in need of further study. The identification of small-molecule ligands that have both allosteric agonist and potentiator activity, termed ago-allosteric modulators, has furthered the number of possible receptor⫺ligand interactions. Such compounds bind to allosteric sites to exert potentiation effects but can also act as agonists in the absence of orthosteric ligand. This agonist effect is often seen at higher concentrations, with the effect transforming into a pure potentiation at decreasing concentrations (Figure 8). Clearly, these receptor⫺ligand-signaling phenomena open the door to a host of speculation about the specific receptor conformations and binding dynamics associated with ago-allosteric modulators, and recent publications address these issues in more depth (22, 23). In addition to PAMs and NAMs, allosteric binding sites on GPCRs allow other novel modes of receptor modulation. Allosteric ligands can display a phenomenon referred to as neutral cooperativity, also coined pharmacological silence. These ligands do not activate
9800 10 µM allosteric 8600 modulator 7400 or vehicle
Add EC20 agonist
EC20 agonist
6200 5000 3800 2600 1400 200 −1000 0
1
2
3
4
5
6
7
Time (min)
Figure 6. HTS screen for a positive allosteric modulator employing calcium fluorescence. To detect potentiation in the screening assay, test compounds were added after 10 s of baseline determination. Five minutes later, a fixed EC20 concentration of orthosteric agonist was added. An “active” was identified as a compound that caused no response in the absence of the orthosteric agonist but a significant (>2-fold) increase in the response to the EC20 concentration of orthosteric agonist.
or inactivate the GPCR in the presence or absence of orthosteric agonist but block the activity of both PAMs ands NAMs by occupying the allosteric site. The first example of this was described for allosteric ligands of mGluR5 within the DFB (14) series (Figure 5) (24). Conn et al. (25) recently reported on the discovery of a fundamentally new mode of GPCR modulation with the report of mGluR5 allosteric “partial antagonists”. Partial antagonists fully occupy the MPEP (17) binding site on the mGluR5 receptor but only partially block agonist response, resulting in partial mGluR5 inhibition (25). As reKEYWORDS ported by Rodriguez and coAgo-allosteric modulator: An allosteric ligand that functions as both an allosteric modulator workers (23), this effort idenand as an agonist on its own (though the tified three compounds that latter is usually only at higher concentrations. only partially inhibited or had Allosteric agonist: A ligand that is capable of receptor activation on its own by binding to a no functional effects on recognition site that is distinct from the mGluR5 response. Coined orthosteric site. M-5MPEP (20) and Allosteric modulator: A ligand that increases or decreases the action of an orthosteric ligand Br-5MPEPy (21), these com(agonist or antagonist) by binding at an pounds represented the first allosteric site. The modulator may enhance partial antagonists of mGluR5 the affinity and/or efficacy of the orthosteric ligand while exerting no effects on its own. inducing a maximal mGluR5 Allosteric site: A ligand binding site that is inhibition of ⬃50%, along distinct from the orthosteric binding site. In with 5MPEP (22), another truest form, there should be no overlap with the orthosteric binding site. neutral, pharmacologically siOrthosteric site: The binding site of the lent allosteric site ligand endogenous agonist. (Figure 9) (25). VOL.3 NO.9 • 530–541 • 2008
535
60
Normalized fluorescence (percent glutamate maximum)
b Normalized fluorescence (percent glutamate maximum)
a PAM + EC20 agonist
50
EC50 for potentiation = 2.6 ± 0.4 µM (n = 6)
40
30 EC20 agonist 20 alone
3.1 × potentiation
10 PAM alone
0 −9
−7
−8
−6
−5
120 Left-ward shift of agonist response curve 100 80 60
100 µM PAM 10 µM PAM 1 µM PAM Vehicle
40 20 0 −7
−4
log [PAM] (M)
−6
−5
log [orthosteric agonist] (M)
Figure 7. Effect of a PAM in two simple cell-based pharmacological assays. a) A small-molecule PAM potentiates GPCR activation by orthosteric agonist. A range of concentrations of a small-molecule PAM were added to cells after 10 s of baseline determination. Five minutes later, a fixed concentration (⬃EC20 concentration) of agonist (glutamate in this figure) was added, and the Ca2ⴙ response was measured. b) A full CRC of the same small-molecule PAM in the presence of a fixed, low concentration of orthosteric agonist. The small-molecule potentiation of response is manifested as increased agonist sensitivity, that is, the EC50 of the orthosteric agonist is leftward shifted.
Sharma and co-workers (26) recently reported on the SAR of another mGluR5 partial antagonist, 23, identified in a high-throughput screening campaign. The analogs of HTS partial antagonist lead 23 within a small library elucidated a “molecular switch” to modulate pharmacological activity (Figure 10). Lead 23, with an unsubstituted distal phenyl ring, fully occupied the MPEP binding site, possessed an IC50 of 486 nM, but only afforded partial response (29% response, 71% partial antagonism), that is, allosteric partial antagonism. Incorporation of small chemical moieties in the 3-position of the distal phenyl ring, such as a 3-methyl group, delivered 24, a full noncompetitive mGluR5 antagonist (IC50 ⫽ 7.5 nM). When the methyl group is moved from the 3-position to the 4-position as in 25, an efficacious (99% of glutamate max) mGluR5 positive allosteric modulator resulted (EC50 ⫽ 3.3 M, 4.2fold shift). The observation of a conserved molecular switch, accessed by toggling between 3- and 4-substitution on the distal phenyl ring, within this chemical series is unprecedented and once again high-
lights the complexities involved in the optimization and development of allosteric ligands. The discovery of allosteric modulators and the intricate mechanisms underlying their pharmacological properties clearly demanded revision and expansion of classical receptor models. However, even prior to consideration of allosteric modulators, the early two-state receptor models had evolved concurrently with the progressive elucidation of more complex ligand⫺receptor interactions (18, 27). First, the ternary complex model (TCM) arose as an expansion of the simple linear twostate model (Figure 11, panel a), taking into account not only the interaction between a ligand and its receptor but also the active receptor (R*) and G-protein and giving rise to a four-point 2D model (Figure 11, panel b). The extended TCM was then introduced to include the spontaneous activation of a receptor that can interact with a G-protein even in the absence of agonist binding (i.e., agonist-independent spontaneous signaling), producing a six-point 2D model (Figure 11, panel c). Beyond this, the eight-point 3D cubic TCM encompasses the
b Normalized fluorescence (%Max)
a PAM + EC20 agonist
80 60
2.8
40 20
Agonism at higher PAM concentrations
0 −9
−8
−7
−6
PAM alone −5
log [PAM] (M)
Normalized fluorescence (%Max_Glu)
100 120 100
Agonism at higher PAM concentrations
80
10 µM PAM
60
1 µM PAM
40
0.1 µM PAM Vehicle
20 0 −9
−8
−7
−6
−5
−4
log [agonist] (M)
Figure 8. Concentration response curves for a prototypical ago-allosteric modulator. a) At low compound concentrations, the molecule behaves like a positive allosteric modulator, eliciting no activation of the receptor in the absence of an EC20 of the orthosteric agonist. Only at higher concentrations is agonism observed. b) In the present case, at concentrations